Skip to main content

An Open Access Journal

Table 2 Feature type, their dimension and description

From: Models for forecasting the traffic flow within the city of Ljubljana

Feature type

Count

Description

One_hot_months

12

Binary vector of the month

One_hot_day_of_week

7

Binary vector of the day of the week

One_hot_minute

96

Binary vector of the 15-minute interval in a day

Sin_cos_day

2

Cyclic time coordinates with a period of one day

Sin_cos_week

2

Cyclic time coordinates with a period of one week

Sin_cos_year

2

Cyclic time coordinates with a period of one year

Linear_year_timespan

1

Time form 2013 to 2020 normalized to [0, 1]

Is_dst_feature

1

Indicator of a daylight saving time

Holiday

1

Indicator of a public holiday

Temperature

1\(^*\)

Average daily temperature

Wind_speed

1\(^*\)

Average daily wind speed

Cloud_coverage

1\(^*\)

Average daily cloud coverage

Humidity

1\(^*\)

Average daily humidity

Air_pressure

1\(^*\)

Average daily air pressure

Rainfall

1\(^*\)

Average daily rainfall

Snowfall

1\(^*\)

Average daily snowfall

Sunshine

1\(^*\)

Average daily sunshine duration

Dew

1\(^*\)

Presence of dew

Glaze_ice

1\(^*\)

Presence of glaze ice as a result of a freezing rain

Icy_ground

1\(^*\)

Presence of ice on the ground due to refreezing

  1. Features are divided according to their source. Features on the top are obtained from the timestamp of the measurement, the holiday feature comes from an eternal source and indicates if a given day is a public holiday. Lastly, weather features describe the statistics from the previous few days. Note, \(^*\) indicates that we can include features from a more distant past. If more than one day is included, the number of weather features is subsequently proportional to the number of days that are included